Meningitis The role of climate for prediction Andy

Meningitis: The role of climate for prediction Andy Morse Ph. D. Department of Geography University of Liverpool A. P. Morse@liv. ac. uk Mark Cresswell Ph. D EGS Manchester Metropolitan University

1. 0 Background Meningococcal Meningitis • • • Bacterial meningitis (Neisseria meningitidis) causes epidemics 12 serotypes are know only 4 cause epidemics A, B, C and W 135 Group A generally causes epidemics in Africa although cases due to serogroups C, X and W 135 are found. • B and C are more common in the U. K. • Vaccines exist for A, C, X, Y and W 135

1. 1 Background Meningococcal Meningitis • Transmitted person to person (sneezing, coughing, kissing) (military recruits, students) • Average period of incubation 4 days ( 2 to 10 days) • Estimated 10 to 25% carry the bacterial but can increase in epidemics • U. K. matter of education and seeking treatment

1. 2 Background Meningococcal Meningitis in Africa • • Meningitis epidemic disease, highly seasonal - later half dry season Epidemics every 5 to 10 years – kills young adults as well as children Climatic connections are ‘not proven’ low humidity (vapour pressure) and dust important factors Epidemics cease with the onset of the rains Figure from Cheesbrough, JS, Morse AP, Green SDR. Meningococcal meningitis and carriage in western Zaire: a hypoendemic zone related to climate? Epidemiology and Infection 1995: 114; 75 -92

1. 3 Background West African Climate • • Area dominated by seasonal rains produced by a monsoonal system Strong latitudinal gradient in ‘wetness’ and thus climates and vegetation Monsoon system is complex and not well understood Leads to large interannual climate

1. 4 Background West Africa Atlas

1. 5 Background West African Climate • Monsoon System and AMMA experiments

1. 6 Background West African Climate NDVI February NDVI August From MARA eshaw website http: //www. mara. org. za/eshaw. htm

1. 7 Background West African Climate Look at Hutchinson rainfall climate maps in unit folder Animation from University of Liverpool Understanding Epidemics Website http: //www. liv. ac. uk/geography/research _projects/epidemics/MAL_intro. htm Data from CLIVAR VACS Africa Climate Atlas at University of Oxford

1. 10 Background Epidemic Cycles • • • Many infectious diseases, in the tropics, have a strong seasonal cycle related to the seasonal climatic cycles Climatically anomalous years can lead to epidemics Time between trigger threshold to epidemic peak often too short to take effective intervention – need for skilful and timely seasonal climate forecast Vaccine Threshold Effect

2. 0 Linking climate to disease Example for meningitis in Africa • Extensive literature search was Spatial Distribution Meningitis Epidemics 1841 -1999 (n = c. 425) 1 undertaken to identify reported epidemics • Published and grey literature were consulted 1 Molesworth A. M. , Thomson M. C. , Connor S. J. , Cresswell M. P. , Morse A. P. , Shears P. , Hart C. A. , Cuevas L. E. (2002) Where is the Meningitis Belt? , Transactions of the Royal Society of Hygiene and Tropical Medicine, 96, 242 -249.

2. 1 Linking climate to disease Example for meningitis in Africa • Statistical Model to produce a map of risk • Epidemiological data and climatic and environmental variables • • Risk factors: Land cover type and seasonal absolute humidity profile • Seasonal dust profile, Population density, Soil type Significant but not included Molesworth, A. M. , Cuevas, L. E. , Connor, S. J. , Morse A. P. , in final model • • Human factors not included Thomson, M. C. (2003). Environmental risk and meningitis epidemics in Africa, Emerging Infectious Diseases, 9 (10), 12871293.

2. 2 Linking climate to disease Example for meningitis in Africa • Cluster analysis to define areas with common seasonal cycle • Absolute humidity values • Used to produce risk map shown above Molesworth, A. M. , Cuevas, L. E. , Connor, S. J. , Morse A. P. , Thomson, M. C. (2003). Environmental risk and meningitis epidemics in Africa, Emerging Infectious Diseases, 9 (10), 1287 -1293.

2. 4 Linking climate to disease Values to give an absolute humidity of about 10 gm-3 T (temperature celsius) T dew (celsius) e (vapour pressure h. Pa) 40 12. 5 14. 5 30 12 14 30 11. 5 13. 6 10 11 13. 1

2. 6 Linking climate to disease Example for meningitis in Africa • Disease is complex and dry air and dust are not the only factors • Many human ones – immunity, nutrition and co-infection • However the environmental variables may lead to the population becoming more susceptible • The environmental variables may be predictable months in advance.

3. 0 Potential of Seasonal Forecasting Background applications • Probabilistic forecasts are made routinely • Statistical models – more established – more regionally and single variable orientated – cannot work outside their training data – can work well e. g. spring SST to summer rains (West Africa) • Dynamic models – Ensemble Prediction Systems – experimental also operational too Loaded dice example – loading and hence predictability changes with time and location •

3. 1 Potential of Seasonal Forecasting Dynamic EPS products • Typical Products from ECMWF

3. 2 Potential of Seasonal Forecasting Dynamic EPS products • Typical Products from ECMWF Probabilistic Seasonal 2 to 4 month lead time

3. 3 Potential of Seasonal Forecasting Combined products International Research Institute for Climate Prediction (IRI), Columbia University, New York Seasonal Forecast 2 to 4 month lead time

3. 4 Potential of Seasonal Forecasting Dynamic EPS – issues for users and producers • • • Tailored verification Verification of user parameters Scale – downscaling Bias correction Weighting Application model and method development – run with EPS Product derived time scale cut off – medium, monthly, seasonal and beyond Interdisciplinary nature of research Taking of academic risk

3. 5 Potential of Seasonal Forecasting Product Verification Met. Office Seasonal Forecast Precip. AMJ 2 to 4 month lead time yellow through red - increasing predictive skill white through dark blue - little or no better than guesswork Units = Gerrity skill score

3. 8 Potential of Seasonal Forecasting Liverpool Malaria Model – LMM • • • Dynamic model Daily time step Driven by temperature and precipitation Observations, reanalysis, ensemble prediction systems Developed within a probabilistic forecasting system – DEMETER Continuing in EMSEMBLES • Model details Hoshen, M. B. and Morse, A. P. (2004) A weather-driven model of malaria transmission, Malaria Journal, 3: 32 (6 th September 2004) doi: 10. 1186/1475 -2875 -3 -32 (14 pages) • Applied in an EPS in Morse, A. P. , Doblas-Reyes, F. , Hoshen, M. B. , Hagedorn, R. and Palmer, T. N. (2005). A forecast quality assessment of an end-to-end probabilistic multi-model seasonal forecast system using a malaria model, Tellus A, 57 (3), 464 -475

4. 0 Summary The Forecasting Triangle Providers Users Dissemination Forecasts Demand Feedback Training + Product Guidance and Development Developers with users and providers

4. 1 Summary • Probabilistic (and deterministic) forecasts are routinely produced operationally leads times days to seasons • This potential resource is under utilised by application user communitiesgaps in knowledge and awareness issues with forecast skill and guidance in products lack of user application know how and appropriate user application models

4. 3 Summary Current and recent research projects • DEMETER EU FP 5 ENSEMBLES EU FP 6 Addressing development and application of ensemble prediction systems • AMMA-EU FP 6, AMMA-UK NERC, West African monsoon observations, modelling impacts

5. 0 Conclusions Infectious diseases must be modelled to allow use within emerging long range forecast technologies. Much has been done to bridge gaps between forecaster and health user but still many gaps Work is on going and a new ‘epimeteorology’ community is emerging

Websites • WHO meningitis site http: //www. who. int/mediacentre/factsheets/fs 141/en/ • Meningitis Research Foundation http: //www. meningitis. org/ • EU and NERC funded AMMA improve ability to predict the West African Monsoon and its impacts on intra-seasonal to decadal timescales. http: //www. amma-eu. org/ and http: //amma. mediasfrance. org/ EU funded ENSEMBLES probabilistic forecasts of climate variability and climate change over timescales of seasons to centuries and the application and potential impacts of these predictions. http: //www. ensembles-eu. org/ • • Washington, R. , Harrison, M, Conway, D. , Black, E. , Challinor, A. , Grimes, D. , Jones, R. , Morse, A. and Todd, M (2004). African Climate Report - A report commissioned by the UK Government to review African climate science, policy and options for action, DFID/DEFRA, London, December 2004, pp 45 http: //www. defra. gov. uk/environment/climatechange/ccafrica-study/